Quantitative structure-electrochemistry relationship for substituted benzenoids using Levenberg-Marquardt artificial neural network

Abstract

Quantitative structure-property relationship models correlating the half-wave potentials (E _1/2) of the benzenoids and its derivatives were developed using both linear and non-linear modelling approaches. Descriptors calculated from molecular structures alone were used to represent the E _1/2 of the benzenoids. A set of 36 compounds were selected and suitable sets of molecular descriptors were calculated. A genetic algorithm-partial least square (GA-PLS) method was used to select the most appropriate molecular descriptors whilst a linear, quantitative structure-property relationship model was developed; using the selected descriptors, a Levenberg-Marquardt artificial neural network (L-M ANN) was employed for the non-linear model development. The stability and prediction ability of models were validated using leave-group-out cross-validation, external test set and Y-randomization techniques. The described model does not parameters require experimental and potentially provides useful prediction for E _1/2 of new benzenoids derivatives.